Discrepancy in fire names in xBD dataset? about xview2_baseline HOT 5 CLOSED

Sorry for getting to this so late. @kkraoj @yangsiyu007 You're right, I have no idea where Carr went. I'm looking into it now. W.r.t. Socal fires, they correspond to the fires in Ventura County in 2017.

from xview2_baseline.

Just as an update, something's up with Carr. It was never released on our public website for some reason. We're still digging into why that happened and how to release it going forward. Give us a few weeks since it's break and there are some other things going on as well!

from xview2_baseline.

@kkraoj @yangsiyu007 Here's the much awaited conclusion on Carr: due to a breakdown in our dataset tracking, the Carr dataset was never labeled. What does this mean for xBD? Nothing much. All statistics that are shown in the paper are computed from the imagery that is actually present. As of now, we do not have funding to get the Carr dataset labeled.

So sorry about this. It was disappointing to dig into it and discover this as well. Luckily, fires are well-represented in xBD, so I hope this is not a hurdle to your research.

from xview2_baseline.

I should add, we'll correct the manuscript to address this discrepancy.

from xview2_baseline.

I have the same question! Which data files correspond to Carr Wildfire 2018 referenced in the manuscript? I think Carr Wildfire was in northern California, so it doesn't seem to correspond to the "socal fire" images...

from xview2_baseline.